Share this post:
Over the past decades, the aggressive scaling of transistors made on rigid silicon wafers has steadily boosted the performance of personal electronics and supercomputers. For emerging applications like real-time analytics and Internet of Things (IoT), high-performance logic circuits and sensors made on flexible or unconventional substrates are needed in order to enable the true computation at the edge. These are several examples of growing areas where flexible nanomaterials, like carbon nanotubes (CNTs), could offer many appealing advantages over rigid silicon, such as low cost, low power, large-area fabrication or even roll-to-roll production. Although CNTs have been widely considered as superior candidates for flexible electronics due to their high mobility, their practical applications have been limited by the lower performance of flexible CNT thin-film transistors (TFTs) compared to those built on rigid substrates (such as silicon wafer or glass). For example, flexible CNT integrated circuits typically exhibit low-speed operation with logic gate delays of over 1 microseconds. However, this situation could be changed with the new advances in IBM Research.
Flexible CNT CMOS integrated circuits with sub-10 nanoseconds stage delays. Pictured: A flexible 5-stage CMOS ring oscillator made on a polyimide substrate. (Figures 1b and 4a in “Flexible CMOS integrated circuits based on carbon nanotubes with sub-10 ns stage delays”, published on Nature Electronics.)
In a recent journal article, Flexible CMOS integrated circuits based on carbon nanotubes with sub-10 ns stage delays, published on Nature Electronics, we demonstrate that high-performance CNT TFTs and complementary integrated circuits can be fabricated on flexible substrates. Riding on the decades-long research on carbon electronics at IBM, we have addressed several key challenges in the fabrication of high-performance flexible CNT electronics, including purity and density of semiconducting CNTs, reliable n-type doping technique for complementary logic, as well as process yield and variation on flexible substrates. Overall, the fabricated flexible CNT TFTs have shown state-of-the-art performance, highlighted by the high current densities (>17 mA/mm), large current ON/OFF ratios (>106), small subthreshold slopes (<200 mV/dec), high mobilities (~50 cm2/Vs) and also excellent flexibility—when wrapping on a finger, the flexible TFTs can still work with no performance degradation.
Integrating all the pieces together, we then took one step further to demonstrate high-speed CMOS ring oscillator—a standard benchmark circuit in any logic technology. The functional 5-stage CMOS ring oscillator exhibits stage delays down to only 5.7 nanoseconds, showing nearly 1000X improvement over previous carbon nanotube work. It also represents the fastest flexible ring oscillator ever made with any nanomaterials including CNTs, organic polymers, oxide semiconductors, and nanocrystals. The superior performance and integration-level demonstration here highlight the potential of using CNTs for future applications such as IoT, edge computing, flexible displays and sensors, where our work provides a useful approach to build scalable, low-cost, and high-speed flexible electronics.
Integrated flexible pressure sensor with an active matrix of CNT TFTs. Pictured: Current mapping of a flexible CNT pressure sensor resembles the shape of “CNT” word stamps. (Figure 4b in “Large-area high-performance flexible pressure sensor with carbon nanotube active matrix for electronic skin”, published on Nano Letters.)
An example of such applications is presented in another journal article, Large-area high-performance flexible pressure sensor with carbon nanotube active matrix for electronic skin, recently published on Nano Letters. In this work, an integrated flexible pressure sensor is demonstrated with an active matrix of 16×16 CNT TFTs to mimic the tactile pressure sensing functionality of human skin. The fully integrated flexible pressure sensor can operate within a small voltage range of 3 V, and shows superb performance featuring high spatial resolution of 4 mm, faster response than human skin (<30 milliseconds), and excellent accuracy in sensing complex objects on both flat and curved surfaces. We are hopeful that our work may pave the road for future integration of high-performance electronic skin in smart robotics and prosthetic solutions.
About the author
Dr. Jianshi Tang earned his PhD degree in Electrical Engineering at University of California, Los Angeles, where he studied the device and physics of various low-dimensional nanomaterials, such as semiconductor nanowires, topological insulators, and magnetic nanostructures. After that, he joined the IBM Thomas J. Watson Research Center in 2015 as a Postdoctoral Researcher, and later was promoted to Research Staff Member, to further pursue his dream of developing nanomaterials and nanoelectronics into viable technologies that can be potentially adopted in semiconductor industries. His current work at IBM involves developing high-performance carbon nanotubes electronics and also exploring various hardware approaches to achieve energy-efficient neuromorphic computing.